mutation_exp

[Instructions]

[mutation_exp] : Find genes with mutations that have statistically different expression of the selected gene.

Samples are divided into two groups. Group A contains a mutation in the selected gene. Group B does not contain a mutation of the selected gene. A t-test is performed for all genes in the chosen expression dataset between samples in Group A vs. Group B. A ranked list of genes is returned.

Example Query: What gene mutations are associated with expression level of EGFR in TCGA-LUAD-RSEM?

Choose muation_exp in pulldown. Select TCGA-LUAD for mutation dataset. Select TCGA-LUAD-RSEM for expression dataset. Enter "EGFR" into hugo box. Press Submit.

Red colored genes have higher EGFR expression when the labelled gene is mutated. Blue colored genes have lower EGFR expression when the labelled gene is mutated.

Box Plots showing the difference in expression level for EGFR between mutated and non-mutated samples for the top six differentially expressed genes by t-test.

Development: Choose the statistical method to calculate p-value. Option is under expression_mutation.
t-test (default): divide samples into two groups based on presence of mutation. t-test expression levels between group and plot the t-stat vs. p-value on volcano plot.
glm : logistic regression - fit a glm model based on mutation status. plot the p-value vs. AIC on volcano plot. ( this method seems to distinguish better between large proteins with many mutations. eg. TTN)

Note: Some samples in TCGA were profiled for expression more than once. Instead of picking one, the mutation will be tested vs. all replicates and may result in over-representation in the output.